CN105046406A - Inpatient medical management quality assessment method - Google Patents
Inpatient medical management quality assessment method Download PDFInfo
- Publication number
- CN105046406A CN105046406A CN201510357957.6A CN201510357957A CN105046406A CN 105046406 A CN105046406 A CN 105046406A CN 201510357957 A CN201510357957 A CN 201510357957A CN 105046406 A CN105046406 A CN 105046406A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- variable
- hospital
- complication
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses an inpatient medical management quality assessment method which screens historical data and builds models, including authenticating and clearing data, DRG(Diagnosis Related Group)and model classifying, classification and collection of ICD complication and other variables during hospital admission, the statistic tests and screening of hospital admission and complication variables, the building of statistic models and quality assessment. The method also screens current data and calculates a predicated value, including predicated value of risk when a patient is sent in hospital, to realize risk predictions on the mortality rate of each patient, the days spent in hospital and the medical cost in hospital as well. According to the invention, the method effectively converts medical data to find solutions so as to achieve the values of data based on the analysis of big data, the mathematic statistical method and machine learning method. The method also makes medical data comparable. The method not only performs assessment to medical quality among diseases but also realizes proper assessment to managements of treatment in a hospital between doctors, between different departments and between different patients.
Description
Technical field
The present invention relates to lean hospital quality management field, the large data analysis of medical treatment and Decision-making of Hospital Management and support field, particularly relate to a kind of inpatient's medical control method for evaluating quality.
Background technology
In recent years due to the fast development of domestic hospitals IT technology, preliminarily complete the raw data accumulation of patient and disease, but suffer from and there is no methodology, these data can not be refined the decision-making foundation becoming tutorial message and hospital management effectively, cause most data can only be stored in the data warehouse of hospital, waste resource.If U.S. government can be used for reference fully to the successful pattern of hospital management and outstanding methodology, in addition localization improvement again, domestic HMOs not only can be allowed to increase effective monitoring approach and means, and hospital can also be impelled to accelerate from the extensive paces made the transition to fine management model.
Recent government actively advocates and encourages traditional industries to the transition of " internet+", make full use of the quality of medical care that digital technology improves hospital, operation efficiency and the waste of minimizing medical resource become the tendency of the day, seize the opportunity, use for reference advanced experience, Criterion pattern, will occupy first-strike advantage, leads the reform tide of industry.
Clinical medical multidisciplinary and complicacy that is disease adds data depth analysis and purifies as the difficulty of management decision-making support foundation, compared with the data of other industry, medical data has nonadditivity (as financial data) and non-immediate comparability (as size of data) feature, due to each hospital admissions patient crowd and disease degree difference, be irrational by directly adopting the data such as mortality ratio, length of stay and medical treatment cost to disease kind, doctor, performance comparative assessment between section office and hospital.For example owing to accepting transfer from one hospital to another patient in a large number and accept the more serious patient crowd of the state of an illness for medical treatment, Sichuan West China Hospital just directly can not carry out simple performance evaluation with certain County Hospital.
In order to effectively solve the predicament of clinical data injustice, the disease group inductive method that it is standard that one of evaluation profile that hospital adopts usually has with resource use, as all kinds of DRG and DCG etc., then the medical treatment cost will used in treatment, by analysis, the case complexity index method (CMI) of disease group is obtained.To be retrodicted out by medical resource service condition the severity extent of Hospital Disease group, thus realize the assessment in same system of hospital and section office.But in evaluation quality of medical care, operation efficiency and Rational drugs use etc., having its congenital deficiency with the method that CMI calculates, first this pattern also reckons without the characteristic of disease itself and other clinical correlation influence factors, does not meet medical rule; Secondly excessive imaging and treatment and the virtual height cost treatment itself that causes also can increase model instability, thus cause the deviation of judged result.
Summary of the invention
The object of the invention is to overcome the deficiencies in the prior art, a kind of novel inpatient's medical control method for evaluating quality is provided, be adjusted to basis with inpatient's disease risks and realize the assessment of full disease hospital quality management, model is by the historical data of all In-patients of certain hospital or a certain area, by complication/complication adjoint during patient admission, individual patient speciality is (as sex, age, survival condition etc.), and state source etc. of being admitted to hospital is integrated into the variation factor of disease treatment, by the treatment information that disease associated group (DRG) classification is final with these patients, set up mortality ratio respectively, the statistics correlativity regression model of length of stay and inpatient medical cost.And then by the algorithm that these models draw, the existing patient of hospital is precisely predicted, calculate the desired value of each patient at mortality ratio, length of stay and inpatient medical cost.
The object of the invention is to be achieved through the following technical solutions: inpatient's medical control method for evaluating quality, comprise a historical data screening and screen and pre-value calculation procedure with modeling procedure, a current data:
Described historical data screening comprises the following steps with modeling procedure:
S1: import historic discharged patient's data from hospital database;
S2: data are differentiated and cleaning, filter out bad data and extreme value data and are deleted;
S3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set;
Each patient has a medical diagnosis on disease associated packets DRG, is realized classification and the assessment of correlativity diagnosis by DRG;
According to DRG, sorted out by the DRG be associated, each DRG is incorporated into a pattern number, is realized classification and the assessment of correlativity DRG by pattern number;
S4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof;
International Disease Diagnosis Standard classification according to human organ and system carries out classification process to the complication of the past during patient admission and complication;
Adopt international disease to close the collective standard of complication, cluster set is carried out to the ICD diagnosis of patient in same relevant disease group DRG or Operation encoding, is formed and close complication class variable;
Its dependent variable comprises age, sex, social economic environment, situation of being admitted to hospital and source-information;
S5: in same DRG group, utilizes Statistical Identifying Method being admitted to hospital and closing complication variable and carry out pre-service appreciable impact patient death rate, length of stay and cost;
S6: set up statistical models, Mortality data adopts Logic Regression Models, length of stay and medical treatment cost data acquisition multiple linear regression model, the Variable Selection of Corpus--based Method LASSO method is used in modeling, and in conjunction with clinical experience analysis, then obtain the distinguished variable coefficient in a model chosen, form the quantitative formula of predicted value;
S7: model quality is verified: adopt the inspection of the C-Index in statistics and the R-square method of inspection to calculate in sample population and non-sample crowd model, evaluate according to corresponding result;
Described current data screening comprises the following steps with pre-value calculation procedure:
SS1: import current discharged patient's data from hospital database;
SS2: data are differentiated and cleaning, filter out bad data and are deleted;
SS3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set;
Each patient has a medical diagnosis on disease associated packets DRG, is realized classification and the assessment of correlativity diagnosis by DRG;
According to DRG, sorted out by the DRG be associated, each DRG is incorporated into a pattern number, is realized classification and the assessment of correlativity DRG by pattern number;
SS4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof;
International Disease Diagnosis Standard classification according to human organ and system carries out classification process to the complication of the past during patient admission and complication;
Adopt international disease to close the collective standard of complication, cluster set is carried out to the ICD diagnosis of patient in same relevant disease group DRG or Operation encoding, is formed and close complication class variable;
Its dependent variable comprises age, sex, social economic environment, situation of being admitted to hospital and source-information;
SS5: the predicted value calculating patient admission risk, realizes the be admitted to hospital risk profile of each inpatient in mortality ratio, length of stay and medical treatment cost; The risk profile of disease refers to find out the universal law of the final treatment results of impact and can quantization factor by planting in the historical data of management in each disease of hospital, infers the prediction occurrence value the current mortality ratio, length of stay and the medical treatment cost that have a patient of similar disease degree and similar features;
The algorithmic formula of predicted value is as follows:
Expected mortality
wherein, b
irepresent significant correlation property coefficient, b
0represent model intercept, n represents the significant correlation variable number of patient;
Length of stay and medical treatment cost
wherein, b
0represent model intercept, MSE represents the square error of model, b
irepresent significant correlation property coefficient, 0.5 is statistic bias modified value;
Final employing reality occurs and expection relative value is assessed inpatient's medical control quality.
Statistical Identifying Method being admitted to hospital and closing complication variable and carry out pretreated step and comprise following sub-step appreciable impact patient death rate, length of stay and medical treatment cost is utilized described in step S5:
(1) statistical test: filter out to the variable that mortality ratio, length of stay and medical treatment cost have a significant impact in each model, conspicuousness variation refers to the Variable Factors utilizing statistical test to select to have result obvious influence;
(2) the variable process of strong correlation connection property: strong correlation connection property variable refers to that two or more variablees are at statistical significance with have strong correlation clinically in a model, if there is the occurrences of High relevancy in certain model, according to demand and the clinical experience judgement of model, between the variable with High relevancy, select the variable larger on model result impact.
The step setting up statistical models described in step S6 comprises the following steps:
(1) list of LASSO generation forecast variable importance is utilized: due to the scaling coefficient that LASSO is exclusive, at scaling coefficient from 0 to the process of maximal value, add up the number of times that each predictive variable occurs in a model, carry out sorted lists according to predictive variable occurrence number, obtain the list of predictive variable importance; The list of predictive variable importance can be reflected in linear regression from an aspect, and each variable affects size to independent variable, and wherein impact is the variable of 0 is the variable just directly deleted after using LASSO;
(2) in conjunction with the critical value of clinical experience determination importance list:
The critical value of mortality model: if predictive variable is the conjunction complication variable of acute illness class, and continuous three variablees are chronic disease class and/or the conjunction complication with present illness dereferenced below, be judged as three variablees next, and all variablees afterwards cause dead impact to be ignored on patient's present illness, delete from first variable after the conjunction complication variable of acute illness class;
The critical value of length of stay and medical treatment cost model: no matter be acute or the conjunction complication variable of chronic disease class, all can produce strong and weak different impacts, so critical value is defined as 0 to treating the resource used;
(3) selection of predictive variable is determined: after predictive variable selection is determined, these variablees can be used to re-establish linear regression model (LRM) with the method for LASSO, here the method used readjusts the coefficient of original all pretreated predictive variables, in importance, be judged to be that its coefficient of unessential variable is forced to be set to 0, thus exclude the possibility entering final mask; In addition, other choose the coefficient range of variable also can specify, reach the strict object controlling final mask quality, and the regulation principle of coefficient range is that the coefficient symbols of variable must be consistent with two standards:
Article 1, standard: coefficient symbols must be consistent with the symbol of statistical test amount in pre-service;
Article 2 standard: coefficient symbols must be consistent with the result of clinical judgment;
The clinical meaning of Article 2 standard pin to variable reality specifies;
(4) regression model is set up: the method for the Optimum utilization Cross-Validation of scaling parameter in model, the standard of Optimal Parameters is according to dtd-data type definition: discrete independent variable uses MisclassificationError, and independent variable uses MeanSquaredError continuously, the parameter making error rate minimum elects the scaling parameter that final mask uses as, and modeling completes.
Model quality verification step described in step S7 is introduced external data and is verified, in disease risks model, modelling verification utilizes independently patient data in the recent period to verify, comprises the following steps:
(1) basic comparative analysis: adopt and the compare of analysis with class model, after the data of same test sample book being inputted two models, classification is realized to result and compare;
(2) statistical testing of business cycles method: in Logic Regression Models, computation model prediction is got over close to 1 be harmonious coefficient C-Index, the C-Index of actual value, and model prediction is better; In linear regression model (LRM), the fitting coefficient R-square of computation model prediction and actual value, R-square is larger, and model prediction is better;
(3) test in test data, test data is independent a data, or the data of the current patient under the prerequisite not having essence to change in other conditions;
(4) carry out C and R by same test sample book data and other with class model to check, compare of analysis model quality.
The invention has the beneficial effects as follows:
(1) Introduced From Abroad advanced management experience and localization are integrated, reformed and improved, define specialty, science, practical management assessment method, this assessment and analysis system based on the large data analysis foundation of hospital's this medical treatment of bulk sample can take into full account the influence degree of disease risks etc., achieve quality of medical care, hospital efficiency, the comprehensive assessment of medical treatment cost control and patient satisfaction, medical quality in hospital management level can be promoted in Comprehensive ground, promote medical lean development and for comprehensive hospital assess, science is located, Priority Department is set up, performance evaluation and comprehensive hospital general management provide scientific basis and reference.
Hospital's disease control intellectual analysis and evaluating system use for reference international advanced medical control experience, and with informationization as carrying, science, precisely, comprehensive assessment medical quality managent, efficiency of operation, medical treatment cost control and patient satisfaction.As the quantification tool of quality of medical care lean management, medical control will be promoted to lean management future development, lead hospital's lean management new direction.
DMIAES efficiently solves the not comparable difficult problem of quality of medical care in hospital management, for administration office of the hospital provides medical quality managent evaluation criteria and decision-making foundation, for hospital's scientific management, performance evaluation provide standard and judgment.
Medical treatment cost and cost model in hospital's disease control intellectual analysis and evaluating system, full disease disease cost accounting and fee calculating are realized, and take into full account the risk factors of various disease, scientific forecasting cost of illness and expense, effective control medical resource waste, plants applying of medical treatment paying for the DRGs disease of dividing into groups by disease and provides reference frame.
(2) be adjusted to basis with inpatient's disease risks and achieve the assessment of full disease control, model is by the historical data of all In-patients of certain hospital or a certain area, by complication/complication adjoint during patient admission, individual patient speciality is (as sex, age, survival condition etc.), and state source etc. of being admitted to hospital is integrated into the variation factor of disease treatment, by the treatment information that disease associated group (DRG) classification is final with these patients, set up mortality ratio respectively, the statistics correlativity regression model of length of stay and inpatient medical cost.And then by the algorithm that these models draw, the existing patient of hospital is precisely predicted, calculate each patient in mortality ratio, the desired value of length of stay and inpatient medical cost is adjusted to basis with inpatient's disease risks and realizes the assessment of full disease control, model is by the historical data of all In-patients of certain hospital or a certain area, by complication/complication adjoint during patient admission, individual patient speciality is (as sex, age, survival condition etc.), and state source etc. of being admitted to hospital is integrated into the variation factor of disease treatment, by the treatment information that disease associated group (DRG) classification is final with these patients, set up mortality ratio respectively, the statistics correlativity regression model of length of stay and inpatient medical cost.And then by the algorithm that these models draw, the existing patient of hospital is precisely predicted, the desired value of each patient at mortality ratio, length of stay and inpatient medical cost can be calculated.Achieve the effective conversion of medical data from data to solution by methods such as large data analysis, mathematical statistics and machine learning, achieve data value.
(3) concept of the conjunction complication variable of disease is added, according to the natural law of clinical treatment, the patient previously state of an illness necessarily has very large influence to the treatment results implemented, and meets clinical medicine rule by the disease risks adjustment of closing based on complication.
(4) features such as the identical ill speciality of same ethnic population, and the normal process of medical treatment can both meet the requirement of statistically homoplasy sample, use historical data to carry out modeling and forecasting and also meet mathematical law.Successfully solve the bottleneck of medical data noncomparabilities with the disease risks adjustment being modeled as basis, in the Deng developed country of USA and Europe, progressively achieve extensive promotion and application.
(5) adopt O/E index manner of comparison, the actual occurrence value/desired value of O/E exponential representation, O/E index <1: illustrate that disease risks is high, but case fatality rate, length of stay or medical treatment cost control lower than expection; O/E index >1: illustrate that disease risks is low, but case fatality rate, length of stay or medical treatment cost control higher than expection, solve an incomparable difficult problem between medical data, the assess medical quality between disease kind can not only be realized, also can realize the performance Rationality Assessment in inpatient's disease treatment management between doctor, between hospital department, between hospital, be that health authorities is to one of effective regulatory measure of subordinate hospital.
(6) except statistical test, the Risk Adjusted model of the application's model and United States Hospital alliance hospital medical quality managent evaluation system is compared, simultaneously carry out the comparison of O/E index with actual occurrence value, the conclusion of com-parison and analysis be the application's model in the precision of prediction and grade of fit in the sample all higher than same class model.
Accompanying drawing explanation
Fig. 1 is disease risks adjustment model process flow diagram of the present invention.
Embodiment
Below in conjunction with accompanying drawing, technical scheme of the present invention is described in further detail, but protection scope of the present invention is not limited to the following stated.
It is as follows that disease risks controls statistical analysis technique summary:
Patient death rate model:
1. variable pre-service:
Chi-squared checks, P value=0.05;
VIF circular test, MaxVIF<5;
2. statistical modeling:
(1) LASSO method produces variable importance list (1000 scaling parameters), utilizes the critical value merging complication judgment variable importance, retains all variablees being greater than critical value, and test to the coefficient symbols of all variablees.
(2) set up LASSO regression model, utilize Cross-Validation to carry out the optimization of scaling parameter, make optimized parameter minimize MisclassificationError.
(3) optimized parameter is selected to obtain LASSO model.
3. model testing:
Utilize LASSO model to predict new patient data, calculate C-index.
Length of stay and medical treatment cost model:
1. variable pre-service:
Logtransformation is carried out to data value;
T-test checks, p value=0.05;
VIF circular test, MaxVIF<5;
2. statistical modeling:
(1) set up LASSO regression model, utilize Cross-Validation to carry out the optimization of scaling parameter, make optimized parameter minimize MeanSquaredError.
(2) optimized parameter is selected to obtain LASSO model
3. model testing:
Utilize LASSO model to predict new patient data, calculate R-squared.
As shown in Figure 1, inpatient's medical control method for evaluating quality, comprises a historical data screening and screens and pre-value calculation procedure with modeling procedure, a current data:
Described historical data screening comprises the following steps with modeling procedure:
S1: import historic discharged patient's data from hospital database;
S2: data are differentiated and cleaning, filter out bad data and extreme value data and are deleted; Computer programming is adopted to complete the cleaning of bad data and extreme value data.
The definition of bad data: the space data (international statistical classification (ICD coding) etc. as without patient's essential information, the information that enters to leave hospital, diseases and related health problems) 1. in data line; 2. the patient data repeated.
The definition of extreme value data: 1. length of stay is the extreme value patient data of 0 day; 2. the extreme value patient data of length of stay outside the number percent of 99; 3. the extreme value patient data that Direct medical cost of being in hospital is less than 900 yuan; 4. after death remains donations patient data.
S3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set; The commercial DRG code machine of 3M is adopted to realize sorting out set.
The definition of DRG: according to the sorting and grouping of patient disease's diagnosis, operation kind, complication and complication, leave hospital situation, sex and age etc., each patient has a DRG; Be conducive to by DRG the classification and the assessment that realize correlativity diagnosis, reduce diagnosis quantity simultaneously, improve the relevance grade of model prediction;
The definition of category of model: according to DRG, sorted out by the DRG be associated, each DRG is incorporated into a pattern number, is also DMIAES basis DRG, by pattern number/DMIAES basis DRG, is conducive to the classification and the assessment that realize correlativity DRG;
S4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof; Adopt the international medical diagnosis on disease criteria for classification of closing complication, adopt computer programming to complete the classification set of variable.
Close the definition of complication variable: the international Disease Diagnosis Standard classification according to human organ and system carries out classification process (see example one) to the complication of the past during patient admission and complication; By classification process, be conducive to the quantity reducing medical diagnosis on disease/operation variable, improve the degree of stability of model prediction performance;
Adopt international disease to close the collective standard of complication, cluster set is carried out to the ICD diagnosis of patient in same relevant disease group DRG or Operation encoding, is formed and close complication class variable, as shown in the table;
Bacterial endocarditis closes complication group (ICD9 coding)
The definition of its dependent variable: its dependent variable comprises age, sex, social economic environment, the information such as situation and source of being admitted to hospital; Be conducive to comprehensive to patient and admission information etc. factor to take into account in Risk Adjusted model, the assessment of risk of being admitted to hospital is more complete.As shown in the table:
Patient basic population information and state variable (part) of being admitted to hospital
S5: in same DRG group, has the conjunction complication group of statistically significant meaning by statistical test method to patient death rate, length of stay and medical treatment cost and other class variables carry out pre-service.
Pretreated step is as follows:
(1) predictive variable having significant correlation with independent variable is filtered out.The process of screening is man-to-man, have ignored the relation between predictive variable in this step, and only from the impact of single predictive variable on independent variable.
(2) correlativity between computational prediction variable, excludes the variable of strong correlation, ensures that the maximum correlativity of all variablees is less than a critical value.
Operation specifically for different model comprises:
The pre-service of mortality model:
(1) adopt Chi-square Test Chi-squared inspection, assumed value p-value is set as 0.05, for retaining comparatively multivariate, does not consider the multiple testing adjustment multipletestingcorrection of P value;
(2) adopt variance inflation factor VarianceInflationFactor (VIF) as test stone, the critical value of VIF is taken as 5, deletes (referring to example one) the predictive variable of VIF > 5.
The pre-service of length of stay and medical treatment cost model:
(1) original distribution of length of stay and medical treatment cost indefinite (can not be normal distribution substantially) is considered, log-transformation logtransformation is carried out to length of stay and medical treatment cost data, make the data after converting more may meet normal distribution, thus more meet the hypothesis of linear regression model (LRM).
(2) t-test (t inspection) inspection is adopted patient's length of stay of the Liang Ge colony of closing complication and symptom not occurring to occur with or without remarkable difference, assumed value p-value is set as 0.05, for retaining comparatively multivariate, do not consider the multiple testing adjustment multipletestingcorrection of P value.
(3) adopt variance inflation factor VarianceInflationFactor as test stone, the critical value of VIF is taken as 5, deletes the predictive variable of VIF > 5.
VarianceInflationFactor (VIF) process in above model adopts the method for loop iteration to delete, namely the VIF value of each predictive variable relative to other all predictive variables is first calculated, then remove maximum after obtaining all VIF values, recalculate the VIF value of remaining variable again, until the maxVIF < 5 of all variablees.
S6: the foundation carrying out statistical models, Mortality data adopts Logic Regression Models, and length of stay and medical treatment cost data acquisition multiple linear regression model, then obtain the distinguished variable coefficient in a model chosen, and forms the quantitative formula of predicted value.
Close the impact of complication variable at different model: the complication/complication be associated with patient's present illness is two kinds of different conditions of the adjoint property disease of patient, cause state by patient's present illness to change, can have an impact to treatment results again conversely, but result object is different with degree.Complication, compared with complication, has to the patient death that present illness causes the effect more strengthened.But the medical resources such as the length of stay that both all cause present illness and medical treatment cost use effect, but cannot tell strong and weak size simply, needs the acting in conjunction depending on itself and patient's present illness.
Statistics LASSO model
(1) LASSO generation forecast variable importance list (referring to example two) is utilized: due to the scaling coefficient that LASSO is exclusive, scaling coefficient is not from 0 (having scaling) to the process of maximal value (maximum scaling), the number of times that each predictive variable occurs in a model can be added up.Relatively important variable should be able to be more than the number of times of relatively unessential occurrences, no matter such as most important variable generally appears in the size of scaling value in each model.So sorted lists can be carried out according to occurrences number of times: from maximum minimum to occurrence number of occurrence number.This list can be reflected in linear regression from an aspect, and each variable affects size to independent variable.Wherein impact is the variable of 0 (occurrence number is 0) is the variable just directly deleted after using LASSO.
(2) in conjunction with the critical value (variable being less than critical value is deleted) of clinical experience determination importance list:
The critical value of mortality model: if predictive variable is the conjunction complication variable of acute illness class, and continuous three variablees are chronic disease class and/or the conjunction complication with present illness dereferenced below, be judged as three variablees next, and all variablees afterwards cause dead impact to be ignored on patient's present illness.From first variable after the conjunction complication variable of acute illness class (not comprising) delete (see example three).
The critical value of length of stay and medical treatment cost model: no matter be acute or the conjunction complication variable of chronic disease class, all can produce strong and weak different impacts, so critical value is defined as 0 to treating the resource used.
(3) selection of predictive variable is determined: after predictive variable selection is determined, these variablees can be used to re-establish linear regression model (LRM) with the method for LASSO.Here the method used readjusts the coefficient of original all predictive variables (pretreated).In importance, be judged to be that its coefficient of unessential variable is forced to be set to 0, thus exclude the possibility entering final mask.In addition, other choose the coefficient range of variable also can specify, reach the strict object controlling final mask quality.The regulation principle of coefficient range follows the coefficient symbols of variable must be consistent with two standards:
1. coefficient symbols must be consistent with the symbol of statistical test amount in pre-service.
2. coefficient symbols must be consistent with the result of clinical judgment.
The clinical meaning of Article 2 standard pin to variable reality specifies.Symbol (except the special case) general provision such as merging complication variable just (>0) is, namely occurs that symptom can increase dead risk, increases length of stay and cost of being in hospital.The regulation of such symbol also meets the general understanding being combined complication.In addition, such as some case, the risk size of Different age group has consistent understanding (such as older can increase risk and medical treatment cost) clinically, and in this case, the setting of coefficient symbols also should be consistent with clinical understanding.
(4) regression model is set up: the Optimum utilization method of Cross-Validation (being generally 5 times and 10 times) of scaling parameter in model, the standard of Optimal Parameters is according to dtd-data type definition: discrete independent variable uses MisclassificationError, and independent variable uses MeanSquaredError continuously.The parameter making error rate minimum elects the scaling parameter that final mask uses as, and modeling completes.
The risk model of full disease divides three major types, more than totally 800 model unit, cover the ill kind diagnosis of institute, DRG and MDC, the desired value of patient in death, length of stay and medical treatment cost/charge etc. then can be judged exactly according to hospital patient historical data, effectively provide the quantifiable foundation of disease risks degree.
S7: model quality is verified: adopt the inspection of the C-Index in statistics and the R-square method of inspection to calculate in sample population and non-sample crowd model, evaluate according to corresponding result.
(1) basic comparative analysis: adopt and the compare of analysis with class model, after the data of same test sample book being inputted two models, classification is realized to result and compare; Although the selection of variable is different with the mode of modeling, final result still has (see example five chart) of comparability.
(2) statistical testing of business cycles method is adopted: the inspection of Logic Regression Models: the coefficient C-Index that is harmonious of computation model prediction and actual value; Wherein C-Index value is more close to 1, and the prediction effect of model is better.The C-Index > 0.7 of model is required in external comparison model inspection.(the C test value of example 5 1)
The inspection of linear regression model (LRM): the fitting coefficient R-square of computation model prediction and actual value; R-square value is more close to 1, and the prediction effect of model is better.The R-square > 0.05 of model is required in external comparison model inspection.(example 52, the R test value of 3)
(3) test in test data, test data is independent a data, or the data (example 51, the reality/predicted value of 2,3) of the current patient under the prerequisite not having essence to change in other conditions (as therapy approach, means etc.);
(4) carry out C and R by same test sample book data and other with class model to check, compare of analysis model quality (example 51, the test value of 2,3).
Current data screening comprises the following steps with pre-value calculation procedure:
SS1: import current discharged patient's data from hospital database;
SS2: data are differentiated and cleaning, filter out bad data and are deleted;
SS3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set;
SS4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof; International Disease Diagnosis Standard classification according to human organ and system carries out classification process to the complication of the past during patient admission and complication; Its dependent variable comprises age, sex, social economic environment, situation of being admitted to hospital and source-information;
SS1 – SS4 is identical with the step of above S1 – S4.
SS5: the predicted value calculating patient admission risk, realizes the be admitted to hospital risk profile of each inpatient in mortality ratio, length of stay and medical treatment cost;
The definition of risk profile value and establishment condition:
Definition: the risk profile of disease refers to find out the universal law of the final treatment results of impact and can quantization factor by planting in the historical data of management in each disease of hospital, utilizes the methods such as large data analysis, mathematical statistics and machine learning precisely to infer the prediction occurrence value of the current mortality ratio, length of stay and the medical treatment cost that have a patient of similar disease degree and similar features;
Establishment condition: essence change does not occur in modeling and current slot for therapy approach and means and the Medical Treatment Price etc. of diagnosis coding, diagnosis classifying method, disease.
Concrete grammar goes out predicted value (see example four) according to all kinds of formulae discovery.
implementation method case
Example one: the predictive variable correlation of kidney failure mortality model is analyzed
Model #218:DRG682,683,684
Data Source: texas,U.S medical center Herman memorial hospital
Patient's sample number: 2587 leave hospital time 7/1/2004-6/30/2014
The distinguished variable number filtered out: 37
Strong correlation connection variable number: 4
Statistical method: VarianceInflationFactor (VIF)
Note: the detection carrying out strong correlation connection property from 37 variablees filtered out again, lower value is VIF value, and the variable of VIF > 5 is marked.
Example two: kidney failure patient death rate model prediction variable importance list
Model #218:DRG682,683,684
Data Source: texas,U.S medical center Herman memorial hospital
Patient's sample number: 2587 leave hospital time 7/1/2004-6/30/2014
Model variable sum: 280
The distinguished variable number filtered out: 37
Statistical method: LASSO
Note: the numerical value below variable represents variable importance, the higher variable of numerical value is larger on model impact.
Example three: the predictive variable of kidney failure mortality model is selected
Model #218:DRG682,683,684
Data Source: texas,U.S medical center Herman memorial hospital
Patient's sample number: 2587 leave hospital time 7/1/2004-6/30/2014
Model variable sum: 280
The distinguished variable number filtered out: 37
Statistical method: LASSO
Note: combination model and clinical experience, variable rhabdomyolysis, chronic liver disease and variable are afterwards deleted; High relevancy variable-organ built-in pipe is also deleted.
Sick mortality model #22:(patient age >=18 of example four: DMIAES) acute ischemic stroke and use thrombolytic agent companion seriously to close complication (MSDRG61), close complication (MSDRG62), without closing complication (MSDRG63).
Data Source: texas,U.S medical center Herman memorial hospital
Patient's number in modeling sample: 996 sample time 7/1/2004-6/30/2014
Model classification: Logic Regression Models
Degree of fitting in modeling sample: C-Index=0.890
Patient death average expectancy rate in model: 68.4%
The algorithmic formula of predicted value is as follows:
Expected mortality
wherein, b
irepresent significant correlation property coefficient, b
0represent model intercept, n represents the significant correlation variable number of patient;
Length of stay and medical treatment cost
wherein, b
0represent model intercept, MSE represents the square error of model, b
irepresent significant correlation property coefficient, 0.5 is statistic bias modified value;
Be 1.54% without the pre-value of dying of illness of patient during disease variable, have pre-value during multiple disease variable to be raised to 64.14%.
Example four is shown and is adopted disease risks adjustment model to risk profile during two different acute ischemic stroke patient admissions, because the differences such as age of patient, sex, conjunction complication and disease degree cause the different disease risks coefficient of dying of dying of illness.
Example five: the quality verification result of model and compare of analysis
Sick mortality ratio
Model #22: acute ischemic stroke DRG61,62,63
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book patient number: 66 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree C-Index checks: DMIAES model in modeling data: 0.890, DMIAES model in test data: 0.964, U model: 0.933.
Model #328: multiple surgery wound DRG957,958,959
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book number: 212 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree C-Index checks: DMIAES model in modeling data: 0.955, DMIAES model in test data: 0.987, U model: 0.982.
Length of stay
Model #22: acute ischemic stroke DRG61,62,63
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book patient number: 66 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree R-square checks: DMIAES model in modeling data: 0.244, DMIAES model in test data: 0.219, U model: 0.211.
Model #76: cardiac valves and other class cardiothoracic surgery DRG219,220,221
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book patient number: 199 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree R-square checks: DMIAES model in modeling data: 0.256, DMIAES model in test data: 0.261, U model: 0.269.
Direct medical medical treatment cost in hospital
Model #22: acute ischemic stroke DRG61,62,63
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book patient number: 66 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree R-square checks: DMIAES model in modeling data: 0.366, DMIAES model in test data: 0.217, U model: 0.149.
Model #205: diabetes DRG637,638,639
Data Source: texas,U.S medical center Herman memorial hospital
Test sample book patient number: 119 leave hospital time 7/1/2004-6/30/2014
Comparison model: the same class model of our DMIAES model and the U.S. (being called for short U model)
Model-fitting degree R-square checks: DMIAES model in modeling data: 0.383, DMIAES model in test data: 0.227, U model: 0.295.
The above is only the preferred embodiment of the present invention, be to be understood that the present invention is not limited to the form disclosed by this paper, should not regard the eliminating to other embodiments as, and can be used for other combinations various, amendment and environment, and can in contemplated scope described herein, changed by the technology of above-mentioned instruction or association area or knowledge.And the change that those skilled in the art carry out and change do not depart from the spirit and scope of the present invention, then all should in the protection domain of claims of the present invention.
Claims (4)
1. inpatient's medical control method for evaluating quality, is characterized in that, comprises a historical data screening and screens and pre-value calculation procedure with modeling procedure, a current data:
Described historical data screening comprises the following steps with modeling procedure:
S1: import historic discharged patient's data from hospital database;
S2: data are differentiated and cleaning, filter out bad data and extreme value data and are deleted;
S3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set;
Each patient has a medical diagnosis on disease associated packets DRG, is realized classification and the assessment of correlativity diagnosis by DRG;
According to DRG, sorted out by the DRG be associated, each DRG is incorporated into a pattern number, is realized classification and the assessment of correlativity DRG by pattern number;
S4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof;
International Disease Diagnosis Standard classification according to human organ and system carries out classification process to the complication of the past during patient admission and complication;
Adopt international disease to close the collective standard of complication, cluster set is carried out to the ICD diagnosis of patient in same relevant disease group DRG or Operation encoding, is formed and close complication class variable;
Its dependent variable comprises age, sex, social economic environment, situation of being admitted to hospital and source-information;
S5: in same DRG group, utilizes Statistical Identifying Method being admitted to hospital and closing complication variable and carry out pre-service appreciable impact patient death rate, length of stay and medical treatment cost;
S6: set up statistical models, Mortality data adopts Logic Regression Models, length of stay and medical treatment cost data acquisition multiple linear regression model, the Variable Selection of Corpus--based Method LASSO method is used in modeling, and in conjunction with clinical experience analysis, then obtain the distinguished variable coefficient in a model chosen, form the quantitative formula of predicted value;
S7: model quality is verified: adopt the inspection of the C-Index in statistics and the R-square method of inspection to calculate in sample population and non-sample crowd model, evaluate according to corresponding result;
Described current data screening comprises the following steps with pre-value calculation procedure:
SS1: import current discharged patient's data from hospital database;
SS2: data are differentiated and cleaning, filter out bad data and are deleted;
SS3: the classification of medical diagnosis on disease associated packets DRG and model, realizes the classification set to medical diagnosis on disease associated packets, category of model, numbering classification set;
Each patient has a medical diagnosis on disease associated packets DRG, is realized classification and the assessment of correlativity diagnosis by DRG;
According to DRG, sorted out by the DRG be associated, each DRG is incorporated into a pattern number, is realized classification and the assessment of correlativity DRG by pattern number;
SS4: when being admitted to hospital, the international statistical classification ICD of diseases and related health problems closes the classification set of complication and dependent variable thereof, realizes the classification to inpatient's complication and complication and dependent variable thereof;
International Disease Diagnosis Standard classification according to human organ and system carries out classification process to the complication of the past during patient admission and complication;
Adopt international disease to close the collective standard of complication, cluster set is carried out to the ICD diagnosis of patient in same relevant disease group DRG or Operation encoding, is formed and close complication class variable;
Its dependent variable comprises age, sex, social economic environment, situation of being admitted to hospital and source-information;
SS5: the predicted value calculating patient admission risk, realizes the be admitted to hospital risk profile of each inpatient in mortality ratio, length of stay and medical treatment cost; The risk profile of disease refers to find out the universal law of the final treatment results of impact and can quantization factor by planting in the historical data of management in each disease of hospital, infers the prediction occurrence value the current mortality ratio, length of stay and the medical treatment cost that have a patient of similar disease degree and similar features;
The algorithmic formula of predicted value is as follows:
Expected mortality
wherein, b
irepresent significant correlation property coefficient, b
0represent model intercept, n represents the significant correlation variable number of patient;
Length of stay and medical treatment cost
wherein, b
0represent model intercept, MSE represents the square error of model, b
irepresent significant correlation property coefficient, 0.5 is statistic bias modified value;
Final employing reality occurs and expection relative value is assessed inpatient's medical control quality.
2. inpatient's medical control method for evaluating quality according to claim 1, is characterized in that: utilize Statistical Identifying Method being admitted to hospital and closing complication variable and carry out pretreated step and comprise following sub-step appreciable impact patient death rate, length of stay and medical treatment cost described in step S5:
(1) statistical test: filter out to the variable that mortality ratio, length of stay and medical treatment cost have a significant impact in each model, conspicuousness variation refers to the Variable Factors utilizing statistical test to select to have result obvious influence;
(2) the variable process of strong correlation connection property: strong correlation connection property variable refers to that two or more variablees are at statistical significance with have strong correlation clinically in a model, if there is the occurrences of High relevancy in certain model, according to demand and the clinical experience judgement of model, between the variable with High relevancy, select the variable larger on model result impact.
3. inpatient's medical control method for evaluating quality according to claim 1, is characterized in that: the step setting up statistical models described in step S6 comprises the following steps:
(1) list of LASSO generation forecast variable importance is utilized;
(2) critical in conjunction with the list of clinical experience determination importance;
(3) selection of predictive variable is determined;
(4) regression model is set up.
4. inpatient's medical control method for evaluating quality according to claim 1, it is characterized in that: the model quality verification step described in step S7 is introduced external data and verified, in disease risks model, modelling verification utilizes independently patient data in the recent period to verify, comprises the following steps:
(1) basic comparative analysis: adopt and the compare of analysis with class model, after the data of same test sample book being inputted two models, classification is realized to result and compare;
(2) statistical testing of business cycles method: in Logic Regression Models, computation model prediction is got over close to 1 be harmonious coefficient C-Index, the C-Index of actual value, and model prediction is better; In linear regression model (LRM), the fitting coefficient R-square of computation model prediction and actual value, R-square is larger, and model prediction is better;
(3) test in test data, test data is independent a data, or the data of the current patient under the prerequisite not having essence to change in other conditions;
(4) carry out C and R by same test sample book data and other with class model to check, compare of analysis model quality.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510357957.6A CN105046406A (en) | 2015-06-25 | 2015-06-25 | Inpatient medical management quality assessment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510357957.6A CN105046406A (en) | 2015-06-25 | 2015-06-25 | Inpatient medical management quality assessment method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105046406A true CN105046406A (en) | 2015-11-11 |
Family
ID=54452933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510357957.6A Pending CN105046406A (en) | 2015-06-25 | 2015-06-25 | Inpatient medical management quality assessment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105046406A (en) |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106202955A (en) * | 2016-07-19 | 2016-12-07 | 中电科软件信息服务有限公司 | Diagnosis associated packets method and system based on intellectual coded adaptation |
| CN107145712A (en) * | 2017-04-06 | 2017-09-08 | 广州慧扬信息系统科技有限公司 | The case history statistical analysis system of complication and complication |
| CN107194137A (en) * | 2016-01-31 | 2017-09-22 | 青岛睿帮信息技术有限公司 | A kind of necrotizing enterocolitis classification Forecasting Methodology modeled based on medical data |
| CN107451390A (en) * | 2017-02-22 | 2017-12-08 | Cc和I研究有限公司 | System for predicting acute exacerbations in patients with chronic obstructive pulmonary disease |
| CN107463769A (en) * | 2017-07-11 | 2017-12-12 | 武汉金豆医疗数据科技有限公司 | A kind of medical receipt and payment expense management method and system based on medical diagnosis on disease associated packets |
| CN107799182A (en) * | 2017-10-16 | 2018-03-13 | 陈昕 | The evaluation method and electronic equipment of complication and complication factor of influence |
| CN107958708A (en) * | 2017-12-22 | 2018-04-24 | 北京鑫丰南格科技股份有限公司 | Risk trend appraisal procedure and system after institute |
| CN107967948A (en) * | 2017-12-07 | 2018-04-27 | 泰康保险集团股份有限公司 | Medical big data analysis method and apparatus |
| CN108268987A (en) * | 2016-12-30 | 2018-07-10 | 郑芳田 | The quality of Suresh Kumar product estimates method |
| CN108320788A (en) * | 2017-01-16 | 2018-07-24 | 医渡云（北京）技术有限公司 | Hospital business analysis method and device |
| CN108564991A (en) * | 2018-04-13 | 2018-09-21 | 重庆医科大学附属儿童医院 | Digitization coding case history wrong identification system based on ICD and its recognition methods |
| CN108899068A (en) * | 2018-06-14 | 2018-11-27 | 四川久远银海软件股份有限公司 | A kind of cure of abnormalities time determines method and apparatus |
| CN108898310A (en) * | 2018-06-28 | 2018-11-27 | 上海市医学科学技术情报研究所 | Hospital inspection Cost Estimation System and device based on virtual laboratory |
| CN109615227A (en) * | 2018-12-12 | 2019-04-12 | 泰康保险集团股份有限公司 | Medical institutions' annual appraisal result method, apparatus, medium and electronic equipment |
| CN109801693A (en) * | 2018-12-14 | 2019-05-24 | 平安医疗健康管理股份有限公司 | Medical record group technology and device, terminal and computer readable storage medium |
| CN110051324A (en) * | 2019-03-14 | 2019-07-26 | 深圳大学 | A kind of acute respiratory distress syndrome anticipated mortality method and system |
| CN111383766A (en) * | 2018-12-28 | 2020-07-07 | 中山大学肿瘤防治中心 | Computer data processing method, device, medium and electronic equipment |
| CN111489821A (en) * | 2020-03-31 | 2020-08-04 | 宜昌市中心人民医院(三峡大学第一临床医学院、三峡大学附属中心人民医院) | Diagnostic group management system |
| CN111797902A (en) * | 2020-06-10 | 2020-10-20 | 西安邮电大学 | Medical X-ray film magnification measuring system and method based on image data analysis |
| CN113688854A (en) * | 2020-05-19 | 2021-11-23 | 阿里巴巴集团控股有限公司 | Data processing method and device and computing equipment |
| CN113723674A (en) * | 2021-08-18 | 2021-11-30 | 卫宁健康科技集团股份有限公司 | Medical risk prediction method based on big data correlation |
| CN114974600A (en) * | 2022-03-04 | 2022-08-30 | 中国环球租赁有限公司 | Case grouping method and device, electronic equipment and medium |
| CN115168669A (en) * | 2022-09-07 | 2022-10-11 | 杭州杏林信息科技有限公司 | Infectious disease screening method and device, terminal equipment and medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101178804A (en) * | 2007-12-05 | 2008-05-14 | 卫生部医院管理研究所 | Hospital information system construction method taking medical quality as core |
| CN101554322A (en) * | 2008-04-09 | 2009-10-14 | 陈敦金 | System for estimating state of critically ill patient in obstetrical department |
| CN103201743A (en) * | 2010-11-08 | 2013-07-10 | 皇家飞利浦电子股份有限公司 | Method of continuous prediction of patient severity of illness, mortality, and length of stay |
| CN103488856A (en) * | 2012-06-15 | 2014-01-01 | 国家人口计生委科学技术研究所 | Low birth weight risk prediction system and method for constructing low birth weight risk prediction system |
| CN104063579A (en) * | 2014-05-16 | 2014-09-24 | 上海亿保健康管理有限公司 | Health dynamic prediction method and equipment based on multivariate medical consumption data |
-
2015
- 2015-06-25 CN CN201510357957.6A patent/CN105046406A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101178804A (en) * | 2007-12-05 | 2008-05-14 | 卫生部医院管理研究所 | Hospital information system construction method taking medical quality as core |
| CN101554322A (en) * | 2008-04-09 | 2009-10-14 | 陈敦金 | System for estimating state of critically ill patient in obstetrical department |
| CN103201743A (en) * | 2010-11-08 | 2013-07-10 | 皇家飞利浦电子股份有限公司 | Method of continuous prediction of patient severity of illness, mortality, and length of stay |
| CN103488856A (en) * | 2012-06-15 | 2014-01-01 | 国家人口计生委科学技术研究所 | Low birth weight risk prediction system and method for constructing low birth weight risk prediction system |
| CN104063579A (en) * | 2014-05-16 | 2014-09-24 | 上海亿保健康管理有限公司 | Health dynamic prediction method and equipment based on multivariate medical consumption data |
Non-Patent Citations (1)
| Title |
|---|
| 易静: ""医院信息数据挖掘及实现技术的探索"", 《中国博士学位论文全文数据库医药卫生科技辑》 * |
Cited By (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194137B (en) * | 2016-01-31 | 2023-05-23 | 北京万灵盘古科技有限公司 | Necrotizing enterocolitis classification prediction method based on medical data modeling |
| CN107194137A (en) * | 2016-01-31 | 2017-09-22 | 青岛睿帮信息技术有限公司 | A kind of necrotizing enterocolitis classification Forecasting Methodology modeled based on medical data |
| CN106202955A (en) * | 2016-07-19 | 2016-12-07 | 中电科软件信息服务有限公司 | Diagnosis associated packets method and system based on intellectual coded adaptation |
| CN106202955B (en) * | 2016-07-19 | 2019-03-01 | 中电科软件信息服务有限公司 | Diagnosis associated packets method and system based on intellectual coded adaptation |
| CN108268987A (en) * | 2016-12-30 | 2018-07-10 | 郑芳田 | The quality of Suresh Kumar product estimates method |
| CN108320788A (en) * | 2017-01-16 | 2018-07-24 | 医渡云（北京）技术有限公司 | Hospital business analysis method and device |
| CN107451390A (en) * | 2017-02-22 | 2017-12-08 | Cc和I研究有限公司 | System for predicting acute exacerbations in patients with chronic obstructive pulmonary disease |
| CN107451390B (en) * | 2017-02-22 | 2020-11-17 | Cc和I研究有限公司 | System for predicting acute exacerbation of chronic obstructive pulmonary disease |
| CN107145712B (en) * | 2017-04-06 | 2021-01-01 | 广州慧扬健康科技有限公司 | Medical record statistical analysis system for complications and complicating diseases |
| CN107145712A (en) * | 2017-04-06 | 2017-09-08 | 广州慧扬信息系统科技有限公司 | The case history statistical analysis system of complication and complication |
| CN107463769A (en) * | 2017-07-11 | 2017-12-12 | 武汉金豆医疗数据科技有限公司 | A kind of medical receipt and payment expense management method and system based on medical diagnosis on disease associated packets |
| CN107799182A (en) * | 2017-10-16 | 2018-03-13 | 陈昕 | The evaluation method and electronic equipment of complication and complication factor of influence |
| CN107967948A (en) * | 2017-12-07 | 2018-04-27 | 泰康保险集团股份有限公司 | Medical big data analysis method and apparatus |
| CN107967948B (en) * | 2017-12-07 | 2021-10-08 | 泰康保险集团股份有限公司 | Medical big data analysis method and device |
| CN107958708A (en) * | 2017-12-22 | 2018-04-24 | 北京鑫丰南格科技股份有限公司 | Risk trend appraisal procedure and system after institute |
| CN108564991A (en) * | 2018-04-13 | 2018-09-21 | 重庆医科大学附属儿童医院 | Digitization coding case history wrong identification system based on ICD and its recognition methods |
| CN108899068A (en) * | 2018-06-14 | 2018-11-27 | 四川久远银海软件股份有限公司 | A kind of cure of abnormalities time determines method and apparatus |
| CN108898310A (en) * | 2018-06-28 | 2018-11-27 | 上海市医学科学技术情报研究所 | Hospital inspection Cost Estimation System and device based on virtual laboratory |
| CN109615227A (en) * | 2018-12-12 | 2019-04-12 | 泰康保险集团股份有限公司 | Medical institutions' annual appraisal result method, apparatus, medium and electronic equipment |
| CN109801693A (en) * | 2018-12-14 | 2019-05-24 | 平安医疗健康管理股份有限公司 | Medical record group technology and device, terminal and computer readable storage medium |
| CN109801693B (en) * | 2018-12-14 | 2023-07-14 | 平安医疗健康管理股份有限公司 | Medical records grouping method and device, terminal and computer readable storage medium |
| CN111383766A (en) * | 2018-12-28 | 2020-07-07 | 中山大学肿瘤防治中心 | Computer data processing method, device, medium and electronic equipment |
| CN110051324A (en) * | 2019-03-14 | 2019-07-26 | 深圳大学 | A kind of acute respiratory distress syndrome anticipated mortality method and system |
| CN110051324B (en) * | 2019-03-14 | 2022-06-10 | 深圳大学 | Method and system for predicting death rate of acute respiratory distress syndrome |
| CN111489821B (en) * | 2020-03-31 | 2021-02-02 | 宜昌市中心人民医院(三峡大学第一临床医学院、三峡大学附属中心人民医院) | Diagnostic group management system |
| CN111489821A (en) * | 2020-03-31 | 2020-08-04 | 宜昌市中心人民医院(三峡大学第一临床医学院、三峡大学附属中心人民医院) | Diagnostic group management system |
| CN113688854A (en) * | 2020-05-19 | 2021-11-23 | 阿里巴巴集团控股有限公司 | Data processing method and device and computing equipment |
| CN111797902A (en) * | 2020-06-10 | 2020-10-20 | 西安邮电大学 | Medical X-ray film magnification measuring system and method based on image data analysis |
| CN111797902B (en) * | 2020-06-10 | 2023-12-22 | 西安邮电大学 | Medical X-ray film magnification measuring system and method based on image data analysis |
| CN113723674A (en) * | 2021-08-18 | 2021-11-30 | 卫宁健康科技集团股份有限公司 | Medical risk prediction method based on big data correlation |
| CN114974600A (en) * | 2022-03-04 | 2022-08-30 | 中国环球租赁有限公司 | Case grouping method and device, electronic equipment and medium |
| CN114974600B (en) * | 2022-03-04 | 2023-09-29 | 中国环球租赁有限公司 | Case grouping method, device, electronic equipment and medium |
| CN115168669A (en) * | 2022-09-07 | 2022-10-11 | 杭州杏林信息科技有限公司 | Infectious disease screening method and device, terminal equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105046406A (en) | Inpatient medical management quality assessment method | |
| CN104992058B (en) | Disease risks adjust method for establishing model | |
| CN104951894A (en) | Intelligent analysis and assessment system for disease management in hospital | |
| Ashfaq et al. | Readmission prediction using deep learning on electronic health records | |
| Liu et al. | A novel approach for failure mode and effects analysis using combination weighting and fuzzy VIKOR method | |
| US20190267141A1 (en) | Patient readmission prediciton tool | |
| Soylu et al. | Multi-criteria inventory classification with reference items | |
| Sadiq et al. | Probability density functions based weights for ordered weighted averaging (OWA) operators: An example of water quality indices | |
| Delen et al. | An analytic approach to better understanding and management of coronary surgeries | |
| Chen et al. | A heterogeneous GRA-CBR-based multi-attribute emergency decision-making model considering weight optimization with dual information correlation | |
| Khaldi et al. | Artificial neural network based approach for blood demand forecasting: Fez transfusion blood center case study | |
| Zolbanin et al. | Processing electronic medical records to improve predictive analytics outcomes for hospital readmissions | |
| Liu et al. | A study on the efficiency of pediatric healthcare services and its influencing factors in China——estimation of a three-stage DEA model based on provincial-level data | |
| Teow et al. | Intelligent analysis of acute bed overflow in a tertiary hospital in Singapore | |
| Lee et al. | An analytical framework for TJR readmission prediction and cost-effective intervention | |
| WO2022160370A1 (en) | Early warning method for novel coronavirus epidemic based on improved triple exponential smoothing model and lstm model | |
| Rahman et al. | Hospital patients’ length of stay prediction: A federated learning approach | |
| Ye et al. | Identifying critical building-oriented features in city-block-level building energy consumption: A data-driven machine learning approach | |
| CN104766127A (en) | Method for predicting the number of hospitalization days of patient after surgery | |
| Duggal et al. | Impact of selected pre-processing techniques on prediction of risk of early readmission for diabetic patients in India | |
| CN109889391A (en) | A kind of network short term traffic forecasting method based on built-up pattern | |
| Westerdijk et al. | Defining care products to finance health care in the Netherlands | |
| Oyebode et al. | Water demand modelling using evolutionary computation techniques: integrating water equity and justice for realization of the sustainable development goals | |
| Jiang et al. | Elite collaborative search algorithm and its application in power generation scheduling optimization of cascade reservoirs | |
| Barbato et al. | On the impact of resource relocation in facing health emergencies |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151111 |
|
| RJ01 | Rejection of invention patent application after publication |